Biologia, Bratislava, 57/Suppl. 11: 59—70, 2002

Comparison of crystal structures of barley α-amylase 1and 2: implications for isozyme differences in stabilityand activity

Xavier Robert1, Richard Haser

1*, Birte Svensson2 & Nushin Aghajari

1

1Laboratoire de Bio-Cristallographie, Institut de Biologie et Chimie des Protéines, UMR 5086-CNRS/UCBL, 7 Passage du Vercors, F-69367 Lyon cedex 07, France; tel.: ++ 33 4 72 72 26 08,fax: ++ 33 4 72 72 26 16, e-mail: r.haser@ibcp.fr2Department of Chemistry, Carlsberg Laboratory, Gamle Carlsberg Vej 10, DK-2500 Copenhagen Valby,Denmark

ROBERT, X., HASER, R., SVENSSON, B. & AGHAJARI, N., Comparison ofcrystal structures of barley α-amylase 1 and 2: implications for isozyme dif-ferences in stability and activity. Biologia, Bratislava, 57/Suppl. 11: 59—70,2002; ISSN 0006-3088.

The germinating barley seed contains two major α-amylase isozyme familiesAMY1 and AMY2 involved in starch degradation to provide energy used bythe plant embryo for growth. The three-dimensional structure of AMY2 hasbeen solved previously, both in the native state and in complex with acarbose,a pseudo-tetrasaccharide acting as a powerful inhibitor for several glycosi-dases. Furthermore, the endogenous bifunctional barley α-amylase/subtilisininhibitor (BASI) present in the mature seeds inhibits both proteases fromthe subtilisin family and AMY2, but not AMY1. The crystal structure of theAMY2-BASI complex has earlier been solved at 1.9 A resolution. In additionto this difference, the isozymes AMY1 and AMY2 show several distinctly dif-ferent properties, despite their high sequence identity. AMY1 and AMY2 thusdiffer in isoelectric point, in affinity for calcium ions, and in stability at acidicpH and elevated temperature. With regard to the enzymatic properties, largevariation is found in the activity towards starch granules and in the affinityfor soluble substrates. The crystal structure of AMY1 has recently been es-tablished at 1.5 A resolution, which allows a direct comparison of the twoisozymes on the basis of the 3D structures. Detailed analysis shows that thestructural organization of the isozymes is virtually identical, and that localchanges are very small. AMY1 and AMY2 thus constitute a remarkable caseof enzymes displaying very similar structures, but important differences inphysico-chemical and enzymatic properties. Here, the recently solved crystalstructure of AMY1 is compared to that of AMY2, with focus on isozyme dif-ferences at areas of interest, i.e. the active site, the starch- granule-bindingsurface site, the area where BASI is recognized by AMY2, and the threecalcium binding sites.

Key words: α-amylase, barley, isozymes, X-ray crystallography, structure com-parison.

* Corresponding author

59

Introduction

α-Amylases (1,4-α-D-glucan glucanohydrolase; EC3.2.1.1) are monomeric enzymes widely occurringin animals, higher plants, and micro-organisms.Their enzymatic function is to catalyse hydrol-ysis of internal α-D-(1,4)-glucosidic linkages instarch (amylose and amylopectin), and relatedoligo- and polysaccharides. The resulting malto-oligosaccharides and glucose are released with re-tained α-anomeric configuration.

In barley seeds, the physiological role of α-amylases is to provide energy to the growing plant-let. The de novo synthesis of the α-amylases, andmore generally of an array of hydrolases, is con-trolled by a complex pathway involving phyto-hormones and launched by external parameterssuch as the temperature and humidity. Two dif-ferent α-amylase isozymes encoded by two multi-gene families can be distinguished in the germinat-ing seeds. These isozymes, known as α-amylase 1(AMY1) and α-amylase 2 (AMY2), contain 414(MW = 45,342 Da) and 403 residues (MW =45,005 Da), respectively (ROGERS & MILLIMAN,1983; ROGERS, 1985a). They show 80% sequenceidentity, but posses low sequence similarity withα-amylases from micro-organisms and animals(ROGERS, 1985b).

AMY1 and AMY2 differ in isoelectric points,and are also known as the low pI and highpI isozyme, respectively (JACOBSEN & HIGGINS,1982). They show significantly different affinity forcalcium ions (BERTOFT et al., 1984; BUSH et al.,1989; RODENBURG et al., 1994), stability at acidicpH (RODENBURG et al., 1994), and stability atelevated temperature. From an enzymatic pointof view, large variation is found in activity andaffinity towards starch granules (MACGREGOR

& BALLANCE, 1980; MACGREGOR & MORGAN,1986) and soluble substrates (SØGAARD & SVENS-SON, 1990; AJANDOUZ et al., 1992). Finally,only AMY2 has the capacity of binding the en-dogenous bifunctional inhibitor BASI (barley α-amylase/subtilisin inhibitor) (MUNDY et al., 1983;SVENDSEN et al., 1986; LEAH & MUNDY, 1989).The high affinity of AMY2 for BASI is indicatedby a Ki of 2.2 × 10−10 M as determined at pH 8and 37◦C (ABE et al., 1993; SIDENIUS et al., 1995),whereas no interaction has been identified betweenBASI and AMY1. Both structural and biochemi-cal data explain these differences (VALLÉE et al.,1998; RODENBURG et al., 2000).

Recently, the three dimensional structure ofAMY1 has been determined to 1.5 A resolu-tion (this manuscript and details in ROBERT, X.,

HASER, R., GOTTSCHALK, T. E., DRIGUEZ, H.,SVENSSON, B. & AGHAJARI, N., to be submitted).The structure of AMY2 has previously been solvedto 2.8 A resolution (KADZIOLA et al., 1994). Here,the native AMY1 structure is briefly described andcompared with that of AMY2.

Material and methods

Crystallization of full-length and truncated AMY1AMY1 (full-length) was first crystallized as previ-ously described (ROBERT et al., 2002) and only two-dimensional, ultra-thin plate-form crystals were ob-tained. These were mechanically unstable due totheir morphology and thus very difficult to handle.Since the failure in obtaining three-dimensional crys-tals of AMY1 is thought to be due to a flexible C-terminus as compared to AMY2, a C-terminal trun-cated form of AMY1, henceforward referred to asAMY1∆9 (ROBERT et al., 2002) was used afterwards.Crystals suitable for X-ray diffraction studies were ob-tained after approximately one month in PEG 8000with 2-methyl-2,4-pentanediol (MPD) as additive and3 weeks using 2-propanol as additive, as previously de-scribed (ROBERT et al., 2002). Crystals leading to thehigh resolution structure of AMY1∆9 were obtainedunder the latter conditions.

Data collection, structure solution, refinementCrystals of full-length AMY1 as well as AMY1∆9belong to the orthorhombic space group P21212. Asearlier described (ROBERT et al., 2002) the molecu-lar replacement method using the AMoRe software(NAVAZA, 2001), as implemented in the CCP4 suite(CCP4 COLLABORATIVE COMPUTATIONAL PROJECT,1994), was used to solve the phase problem of the fulllength, as well as the AMY1∆9, form. The 3D struc-ture of AMY2 (KADZIOLA et al., 1994), without thebound calcium ions and water molecules, was used asa search model. Refinement of the models, alternatingwith manual building, were done using the simulatedannealing routine as implemented in the software CNSversion 1.0 (BRÜNGER et al., 1998). The final R factorfor the high resolution AMY1∆9 structure was 13.6%with an Rfree of 16.3%.

Sequence and structure alignmentThe sequence alignment showing the superimposi-tion of secondary structures and domains betweenAMY1 and AMY2 was rendered using the programESPript (GOUET et al., 1999). Structural superimpo-sition of AMY1 and AMY2 Cα backbones was per-formed with the “rigid body fit” option in TURBO-FRODO (ROUSSEL & CAMBILLAU, 1989). Figuresshowing overall and detailed structures were gener-ated with the programs TURBO-FRODO (ROUSSEL &CAMBILLAU, 1989), GRASP (NICHOLLS et al., 1991),and VIEWERLITETM (freeware from Accelrys Inc.,San Diego, USA – http://www.accelrys.com).

60

Accession numberThe coordinates and structure factors of the AMY1∆9crystal structure solved to 1.5 A resolution have beendeposited at the RCSB (Research Collaboratory forStructural Bioinformatics) Protein Data Bank (PDB –http://www.rcsb.org) under the entry code 1HT6.The coordinates of the native AMY2 structure used inthis comparative study are present in the PDB underthe accession number 1AMY.

Results

Overall structureThe three-dimensional structure of AMY1∆9 con-sists of 404 amino-acid residues, and all side-chainsare clearly identified in the electron density map.It turned out that quite a large number of aminoacid residues (16) throughout the entire polypep-tide display double conformations (ROBERT, X.,HASER, R., GOTTSCHALK, T. E., DRIGUEZ, H.,SVENSSON, B. & AGHAJARI, N., to be submitted).The structure confirms the sequence correctionto AMY1 (as deposited in SWISS-PROT, withentry name AMY1 HORVU and primary acces-sion number P00693): Ala 308 in AMY1 HORVUis in fact a valine (residue 284 in the matureAMY1 numbering) as found by cDNA sequenc-ing and protein mass-spectrometry (SØGAARD etal., 1993a; ANDERSEN et al., 1994). Nine residues(406 to 414) were removed from AMY1 to ob-tain a C-terminal end identical to that of AMY2(ROBERT et al., 2002). Moreover, in the high res-olution AMY1∆9 structure, electron density wasalso lacking for Asn405 (ROBERT, X., HASER, R.,GOTTSCHALK, T.E., DRIGUEZ, H., SVENSSON, B.& AGHAJARI, N., to be submitted). In the crystalstructure of full length AMY1, residues 404 to 414are not visible in the electron density either, andresidues 402 and 403 are not well-defined. Thesefacts support the idea of the C-terminal part ofAMY1 being very flexible or highly disordered,and are consistent with the improvement of crystalgrowth with the C-terminally truncated AMY1∆9form. The structures (full length and AMY1∆9)obtained contain three calcium ions and in the 1.5A resolution AMY1∆9 structure a large number ofwater molecules as well as some molecules of ethy-lene glycol (used as cryo-protectant) were found.

Superimposing the structure of AMY1 onAMY2 (Fig. 1) shows their very high similarity.As observed in almost all α-amylases, AMY1 con-sists of three domains, including a major cen-tral domain (domain A) having a parallel (β/α)8-barrel super-secondary structure. As in AMY2(KADZIOLA et al., 1994), domain A of AMY1 dif-

fers from a classical TIM-barrel eight-fold symme-try (BANNER et al., 1975) due to the presence ofthree additional α-helices in domain A: A-α6a, A-α7a and A-α8a (Fig. 2), as well as of an irregularloop comprising 65 residues (domain B) which pro-trudes from domain A and connects strand β3 tohelix α3. As was described for AMY2 (KADZIOLA

et al., 1994), domain B forms an anti-paralleltwisted β-sheet together with the 26 residues-longprotruding loop, Aβ2-α2, belonging to domain A.The C-terminal domain (domain C) is constitutedof 61 residues and is organized as a five-strandedanti-parallel β-sheet.

The main backbone differences betweenAMY1 and AMY2 are observed within some ofthe barrel loops. Firstly, the loop joining theshort α-helices η3 and η4 (helices having lessthan five residues according to the notation usedin the sequence alignment as presented in Fig-ure 2 – residues 265 to 274, AMY1 number-ing) has a turn induced by Pro268 (Thr266 inAMY2). The 2.4 A deviation between the sum-mit of these loops can be explained by an im-portant sequence difference at the beginning ofthis loop (sequence IDPQ in AMY1, RGTD inAMY2), with a proline (Pro268) being presentonly in AMY1. Due to its rigid nature, thisresidue can promote a slightly different localbackbone folding in AMY1 compared to AMY2.Secondly, a significant deviation in the back-bone was observed in the region located be-tween Val206 and Pro218 (AMY1 numbering),which forms a large loop between β5 and α5,and protrudes from the barrel. The maximumdeviation between Cα’s of AMY1 and AMY2for this loop is 3.4 A and is located at thelevel of the Gly214AMY1 residue (Gly213AMY2).This shift can be explained by the fact thatthe neighbouring region of this residue in AMY2is involved in crystal contacts (residue 209 to214 in AMY2 numbering) as well as in AMY1(residue 212 to 216) – see sequence alignmentFigure 2. Nevertheless, due to different crys-tal packings, symmetry-related molecules involvedin inter-molecular contacts differ in AMY1 andAMY2, in terms of number and strength of in-teractions. In AMY2, inter-molecular interactionsare Tyr211-N/Asp113-Oδ#2 (2.9 A) and Gly213-N/Asp360-Oδ#4 (3.1 A) (KADZIOLA et al., 1994)where #2 and #4 are symmetry related moleculesin AMY2. In AMY1, inter-molecular interactionsare: Thr212-Oγ1/Arg156-Nη1#4 (3.3 A), Gly214-O/Asp118-O#4 (3.3 A), Asp215-N/Asp118-O#4(3.3 A) and Gly216-N/Arg156Nε#4 (4 A) where#4 is a symmetry related molecule in AMY1.

61

Fig. 1. Superimposition of AMY1 on AMY2. AMY1and AMY2 in backbone presentations are shown inred and blue, respectively. The three calcium ions (pre-sented by green spheres) are perfectly superimposablein the two structures.

Fig. 2. Sequence alignment of barley AMY1 and AMY2, calculated with the program CLUSTALW (THOMPSON

et al., 1994) and presented using the program ESPript (GOUET et al., 1999). Identical residues are highlighted inorange. The secondary structure annotation is indicated over the alignment, coloured in blue, green and red fordomains A, B and C respectively. Catalytic residues are highlighted in blue, whereas AMY2 residues involvedin the interaction with BASI are highlighted in red. The asterisks indicate residues interacting with calciumions and are coloured as a function of length of interaction (red, distances less than 3.2 A and black, distancesbetween 3.2 and 4.0 A). Finally, for both enzymes, residues involved in crystallographic contacts are indicatedwith an “a”, the colour code being the same as for the calcium ligand distances.

62

Table 1. Calcium-ion-binding sites in AMY1 and AMY2.

Calcium ions Ligands in AMY1 Distances in AMY1(A)

Ligands in AMY2 Distances in AMY2(A)

Ca 500 Asn92 Oδ1Asp139 Oδ1Asp139 Oδ2Ala142 OAsp149 Oδ2Gly184 OWat630

2.372.492.372.342.352.312.41

Asn91 Oδ1Asp138 Oδ1Asp138 Oδ2Ala141 OAsp148 Oδ2Gly183 OWat630

2.12.32.72.12.62.12.1

Ca 501 Glu109 Oε1Glu109 Oε2Thr112 OAsp114 OAsp118 Oδ1Asp118 Oδ2Wat604Wat656

2.882.412.462.422.612.352.382.30

Glu108 Oε1Glu108 Oε2Thr111 OAsp113 OAsp117 Oδ1Asp117 Oδ2Wat604Wat656

2.92.62.42.22.62.42.22.1

Ca 502 Asp128 Oδ1Asp128 Oδ2Asp143 Oδ1Asp143 Oδ2Phe144 OAla147 OAsp149 Oδ1Wat788

3.582.362.423.322.322.342.342.30

Asp127 Oδ1Asp127 Oδ2Asp142 Oδ1Asp142 Oδ2Phe143 OAla146 OAsp148 Oδ1a

2.62.42.83.42.22.22.5a

a An electron density which could correspond to a water molecule with low occupancy was reported (KADZIOLA

et al., 1994).

Thirdly, a deviation occurs in the vicinityof Gly243 (β4), one of the two residues insertedin AMY1 compared to AMY2 (Fig. 2). Interest-ingly, this insertion does not cause major struc-tural changes in β4, and from Val245 the back-bones of both isozymes are perfectly superimpos-able. Also, the secondary structures are perfectlyconserved around Gly243.

Calcium binding sitesBoth AMY1 and AMY2 (KADZIOLA et al., 1994)structures contain three Ca2+, two of which haveligands only from domain B (Ca501 and Ca502).The third so-called conserved calcium ion is foundin all other α-amylase 3D structures (Ca500), andhas one ligand from domain A and the remain-ing ones from domain B (MATSUURA et al., 1984;BOEL et al., 1990; QIAN et al., 1993; BRAYER etal., 1995; RAMASUBBU et al., 1996; STROBL et al.,1997; AGHAJARI et al., 1998b; MACHIUS et al.,1998; DAUTER et al., 1999). Calcium ions appearto be essential for proper folding, conformationalstability and therefore for activity (BERTOFT etal., 1984; JONES & JACOBSEN, 1991)

Superimposition of the structure of AMY1 on

AMY2 demonstrates both the perfect spatial iden-tity of the three calcium ions and of all the amino-acid and water ligands. These three calcium-ion-binding sites of AMY1 are presented in Figure 3.The only structural difference seen between theisozymes, when comparing the calcium bindingsites, is the presence of an extra water molecule(Wat788) as a ligand in AMY1 for Ca502, but itshould be mentioned that a water molecule withlow occupancy was seen in the Fo-Fc electrondensity map of AMY2, but not included in therefinement (KADZIOLA et al., 1994). In AMY1,Ca500 has 7 interactions with its ligands, oneof these being a water molecule (Wat630). Thishepta-coordination forms a distorted pentagonalbipyramid and is thus similar but not identical tothe homologous site in other α-amylases, whichall are octa-coordinated. Ca501 has 8 ligands in-cluding 2 water molecules (Wat604 and Wat656)and finally, Ca502 is hepta-coordinated including1 water molecule (Wat788) as ligand, determin-ing a pentagonal bipyramid geometry. Interactionsbetween calcium ions and their ligands, both inAMY1 and AMY2, are summarized in Table 1.Two of the calcium ions (Ca500 and Ca502) are

63

Fig. 3. Calcium binding sites in AMY1. Ca2+ ligand interactions are indicated as dot-and-dash lines. (a) is anoverall view of the Ca500 and Ca502 binding sites for which Asp 149 is a ligand in both cases. (b) is a view ofthe Ca501 binding site.

Fig. 4. Close-up of the active site of AMY2 in complex with the acarbose inhibitor (KADZIOLA et al., 1998)showing catalytic residues (labeled in red) along with the remaining substrate-binding residues. Direct hydrogen-bond interactions are shown by dot-and-dash lines. The three sugar rings of acarbose seen in the structure(coloured in yellow) are located in subsites −1, +1 and +2.

very close to each other, and share a ligand (Asp149). The third calcium ion (Ca500) is locatedabout 20 A from the two others. All amino acid

residues interacting with these calcium ions belongto domain B, except Gly184, which is located inloop β4α4 of domain A.

64

Table 2. Active site residues in AMY2 and the corre-sponding residues in AMY1.a

Residues of AMY2 in-volved in the active site

Corresponding residueson AMY1

Tyr51 OHis92 Nε1Arg177 Nη1Asp179 Oδ2Lys182 NζGlu204 Oε2Glu204 Oε1Trp206 OSer208 NSer208 OγHis288 Nε2Asp289 Oδ1Asp289 Oδ2

Tyr52 OHis93 Nε1Arg178 Nη1Asp180 Oδ2Arg183Glu205 Oε2Glu205 Oε1Trp207 OAsn209Asn209His290 Nε2Asp291 Oδ1Asp291 Oδ2

a The trio of catalytic residues is presented in bold.Non-conserved residues between AMY1 and AMY2 areunderlined.

Active siteThe active site of AMY2, identified in the crys-tal structure of a complex between the enzymeand acarbose (a pseudotetrasaccharide resemblinga transition-state analogue) (KADZIOLA et al.,1998), appears to be a large V-shaped depressionon one side of the enzyme, formed by parts of do-main A and domain B. Only two AMY2 residuesinvolved in substrate binding and catalysis dif-fer in AMY1: Lys182/Arg183 and Ser208/Asn209(AMY2/AMY1 numbering); see Table 2. The Cα’sof these residues are very well superimposed.As concerns the side chains of Lys182AMY 2 andArg183AMY1, respectively, they are oriented in op-posite directions. In fact, in AMY2, the presence ofGlu227 (Asn228 in AMY1) does not leave enoughspace for Lys182 to fill the same position as Arg183in AMY1, while in AMY1 Asn228 leaves enoughroom for Arg183.

Lys182 Nζ (AMY2) provides a hydrogenbond to aca1-O3C at subsite +2 (KADZIOLA etal., 1998) (see Fig. 4). When examining the struc-tures, it seems that the AMY1 Arg183, despiteits difference in orientation at this position, doesnot represent a drastic difference since the residuesare of approximately the same length and pos-sess nitrogen atoms capable of forming a hydro-gen bond with an oxygen in acarbose or a sugarsubstrate. A similar conclusion can be drawn forthe couple Ser208/Asn209 (AMY2/AMY1 num-bering). In the AMY2/acarbose complex, one of

the hydrogen bonds between Ser208 and aca1-O2Cat subsite +2 (KADZIOLA et al., 1998) involves thenitrogen of the peptide bond (see Fig. 4). AMY1has this atom at the same location, due to thealmost perfect overlay of the backbone in this re-gion. A second hydrogen bond is formed by Ser208Oγ. Due to the comparable lengths of Ser and Asnand to the presence of an oxygen atom in both sidechains, it seems possible to establish the equiv-alent hydrogen bond in AMY1-substrate and/orsubstrate analogue complexes.

In the AMY2/acarbose complex (KADZIOLA

et al., 1998), a chain of five water molecules wasobserved running from the active site cleft, to-wards the interior of the protein. The first wa-ter molecule in this chain, Wat607, bridges thecarboxylate groups of Glu204 and Asp289, two ofthe three catalytic residues. This water has beenproposed earlier as candidate for participation inthe enzymatic process as the first catalytic watermolecule in hydrolysis of substrates (KADZIOLA

et al., 1994; AGHAJARI et al., 1998b). The pres-ence of these five water molecules, defining a wa-ter pocket, is not correlated with binding of asubstrate, a substrate analogue like acarbose orof the bifunctional α-amylase/subtilisin inhibitor(BASI). Indeed, these water molecules were foundin the native AMY2 structure (KADZIOLA et al.,1994) and in the structure at 1.9 A resolution ofthe AMY2/BASI complex (VALLÉE et al., 1998),whereas a like chain of buried water moleculeslinked to a chloride effector ion was observed inthe psychrophilic α-amylase, AHA (AGHAJARI etal., 1998a). The 3D structure of AMY1 has a waterpocket similar to the one found in AMY2.

The starch granule-binding surface siteIn AMY2, two contiguous tryptophan residues,Trp276 and 277 (Trp278 and 279 in AMY1) con-stitute the so-called starch-granule-binding site(GIBSON & SVENSSON, 1987) on the surface ofthe enzyme (KADZIOLA et al., 1994). This site(Figs 5a,b), which is independent from the ac-tive site, was previously confirmed by UV differ-ence spectroscopy (GIBSON & SVENSSON, 1987;SØGAARD et al., 1993b). Its low binding affinityfor acarbose and β-cyclodextrin, which competeswith starch granules (WESELAKE & HILL, 1983;SØGAARD et al., 1993b), suggests a role in bind-ing of the enzyme to starch in vivo. This site hasbeen proven also by X-ray crystallography in theAMY2/acarbose complex, solved to 2.8 A resolu-tion (KADZIOLA et al., 1998) in which a disaccha-ride unit was stacking to the indole rings of thetwo tryptophans.

65

Fig. 5. Close-up on the starch-granule-binding surface site in AMY2 (a) and AMY1 (b) showing the two adjacenttryptophan residues. The electron density 2Fo-Fc map contoured at 1σ is shown in both cases, for which theresolutions are 2.8 A and 1.5 A, respectively.

Fig. 6. Overall view of AMY1 and AMY2 surfaces coloured as a function of charges where positive and negativecharges are presented in blue and red, respectively. The active site and the BASI recognition area of AMY2,and its corresponding area in AMY1, are highlighted in this figure, where domain B occupies the upper part ofthe molecule, and domain C, the lower part. Representations for AMY1 as well as AMY2 are generated on thesame scale with the program GRASP (NICHOLLS et al., 1991).

Trp278 and Trp279 in AMY1 superimpose re-markably well with the AMY2 counterparts. Eacharomatic side-chain defines a plane and the an-gle between the two planes was determined to beabout 135◦. This angle is preserved in AMY1 andAMY2 in the native state and in the AMY2 com-

plex with acarbose. The conservation of the spatialposition and orientation of the two residues canbe explained in that tryptophans are large bulkyresidues, and that these particular tryptophansare locked by the neighbouring residues: Trp278in AMY1 is situated next to Arg226 (Arg225 in

66

AMY2) and Trp279 next to Lys282 (Lys280 inAMY2). These basic residues, which are held ina fixed position by several hydrogen bonds andhave little space for moving, lock the position ofthe tryptophans and thus this local structure isperfectly conserved between AMY1 and AMY2.

The interface of the protein-protein complexAMY2-BASI and its counterpart in AMY1BASI belongs to the soybean trypsin inhibitorfamily and inhibits AMY2, but not AMY1. Thecrystal structure of AMY2 – BASI was deter-mined at 1.9 A resolution (VALLÉE et al., 1998), inwhich BASI contains a 12-stranded β-barrel struc-ture and inhibits AMY2 by sterically occludingaccess of substrate to the active site. The totalburied surface area in AMY2-BASI is 2355 A2 andcomprises three interacting regions situated in theneighborhood of the catalytic site of AMY2. BASI,however, does not interact directly with the threecatalytic residues, Asp179AMY2, Glu204AMY2 andAsp289AMY2(VALLÉE et al., 1998).

Remarkably, a large cavity, in which a sol-vated calcium ion is trapped, is created at theAMY2-BASI binding interface and stabilized bysix water molecules in a trigonal biprismatic geom-etry. Five water molecules make hydrogen bondsto the three catalytic residues (Asp179, Glu204and Asp289) and to two oxygen atoms of Glu168and Tyr170 from BASI (Tab. 3). Six of eightresidues interacting with BASI by direct hydro-gen bonds are changed in AMY1. The critical roleof some of these residues has been confirmed byprotein engineering (RODENBURG et al., 2000).It should be mentioned that significant sensitiv-ity for BASI could be introduced by mutage-nesis in an insensitive barley α-amylase hybridisozyme. At least three residues of AMY2 arecrucial in the AMY2/BASI interaction (Tab. 3):Arg128AMY2 which makes a charge interactionwith Ser77BASI, Asp142AMY2 forming a salt bridgeto Lys140BASI, and the cis-Pro129AMY2 residuepresumably ensuring an appropriate geometryof Arg128AMY2. In the above-mentioned study,Arg128AMY2 and Asp142AMY2 were mutated toglutamine and asparagine, respectively. A BASI-sensitive isozyme hybrid (AMY1-(1-90)/AMY2-(90-403)) was created with Ki of 0.33 nM, com-pared to 0.22 mM in the case of AMY2. Moreover,Arg128AMY2 was introduced in AMY1 in additionto the Lys130Pro mutation (in order to introducethe equivalent in AMY1 of Pro129AMY2). Thistime, an inhibitor-insensitive hybrid (AMY1-(1-161)/AMY2-(161-403)) was used and theThr129Arg-Lys130Pro mutation conferred sensi-

Table 3. Residues involved in direct hydrogen bonds inthe AMY2/BASI complex and homologues in AMY1.a

Direct hydrogen bonds betweenAMY2 and BASI residues

Correspondingresidues in AMY1

AMY2 residues BASI residues

Arg128 NεAsp142Asp142 OAsp142 Oδ2Gly144 NLys182 NζTyr211 NGly215 OGln223 Oε1His295 OHis295 O

Ser77 OγTyr131 OLys140 NζLys140 NζAsp150 Oδ1Glu168 Oε1Ala28 OTrp162 Nε1Asn26 Nδ2Arg106 NεArg106 Nη1

Thr129Asp143Asp143Asp143Ala145 NArg183 Nη1Thr212 NGly216 OAla224Ala297 OAla297 O

a Non-conserved residues between AMY1 and AMY2are presented in bold.

tivity to BASI corresponding to Ki = 7 µM(RODENBURG et al., 2000). Despite these remark-able mutational alterations of the sensitivity to-wards BASI, it was concluded that the optimalrecognition of BASI depended on several individ-ual enzyme/inhibitor contacts.

Examination of the surfaces of AMY1 andAMY2 (Fig. 6) colored as a function of chargeshows that the BASI recognition area in AMY2is mainly negatively charged, and that the cor-responding region in AMY1 displays a notablywider negatively-charged area. The structural su-perimposition of the two native isozymes andthe examination of residues involved in directcontacts with BASI in AMY2 show that theseresidues and their counterparts in AMY1 can beclassified in two categories (Tab. 3): 1) residuesnot necessarily conserved between AMY1 andAMY2, but which are capable of making the sameinteractions in case a complex between AMY1and BASI could be established. These residuesinclude: Asp143AMY1, which is perfectly super-imposed onto Asp142AMY2; the peptide bondnitrogen atom of Ala145-NAMY1, superimpos-able with Gly144-NAMY2; Thr212-NAMY1 super-imposable on Tyr211-NAMY2; Gly216-OAMY1 andGly215-OAMY2, where the position of the pep-tide bond oxygen atom is conserved, and fi-nally, Ala297-OAMY1 and His295-OAMY2 with thesame kind of conservation; 2) residues that dis-play drastic differences between the two isozymes:Thr129AMY1 not only has a shorter side chain com-

67

pared to its counterpart in AMY2 (Arg128AMY2),but also has different charge properties;Ala224AMY1 compared to Gln223-OεAMY2 is a hy-drophobic, uncharged residue that is unable to in-teract with Asn26-NδBASI.

Finally Arg183AMY1/Lys182-NζAMY2 can beconsidered as having comparable charges and side-chain lengths. One can therefore not conclude thatthis mutation is a severe hindrance to the creationof an interaction with Glu168-Oε1BASI as observedin the AMY2/BASI complex.

A part of the AMY2/BASI complex forma-tion is certainly the solvated calcium ion foundat the interface between the two proteins. Thiscalcium only has water molecules as ligands, fiveof which are interacting with the three catalyticresidues of AMY2. The fact that this calcium ionwas not found in the native structure of AMY1 isfar from being understood, in that the catalyticresidues as well as the environments for the cal-cium ion in both isozymes superimpose perfectly.Furthermore, it is presently difficult to reach a con-clusion on the exact role of this calcium in thenon-inhibition of AMY1 by BASI.

Discussion

The three-dimensional structure of barley AMY1is the second structure of a plant α-amylase solvedto date, the first being that of AMY2 (KADZIOLA

et al., 1994). These two isozymes display conspicu-ous biochemical, biophysical and enzymatic differ-ences. The crystal structure of AMY1 describedhere represents the first step in a study aimingat an improved understanding of these dissimilar-ities. It seems, however, difficult to explain any ofthese differences based on the two native struc-tures. Indeed, the three-dimensional structures ofAMY1 and AMY2 are virtually the same, withthe three domains A, B and C of the two struc-tures perfectly superimposable (Fig. 1), with a fewvariations occurring in specific loops. However, thespatial location of these loops makes it unlikelythat the observed structural deviations are respon-sible for such differences in biochemical properties.For example, the AMY1 and AMY2 contain threecalcium ions with exactly the same environment inboth isozymes; however they display significantlydifferent calcium affinities.

As concerns the differences in stability atacidic pH and at elevated temperature, they mightbe related to subtle structural particularities as isgenerally the case for proteins of different ther-mostability. Thus a thorough structure compari-son including surface electrostatic potential and

dynamic studies is required to address such ques-tions.

As described above, the active sites of AMY1and AMY2 display some differences, but the cat-alytic machinery seems to be identical. How-ever, on the basis of the native structure ofAMY1 and AMY2, it remains difficult to under-stand the large differences in activity and affinityfor starch granules (MACGREGOR & BALLANCE,1980; MACGREGOR & MORGAN, 1986) and affin-ity for soluble substrates (SØGAARD & SVENSSON,1990; AJANDOUZ et al., 1992). The affinity forthese substrates is dictated by the availability,the number and the structure of the subsites.The AMY2/acarbose complex (KADZIOLA et al.,1998) (Fig. 4) shows occupancy in the three sub-sites −1, +1 and +2 (nomenclature according toDAVIES et al., 1997). Moreover, it has been foundby enzymatic subsite mapping (AJANDOUZ et al.,1992) and predicted by computer-aided modelling(ANDRÉ et al., 1999), that both AMY2 and AMY1possess 10 subsites, namely −6 through −1 to-wards the non-reducing end from the site of cleav-age and subsites +1 through +4 towards the re-ducing end of the substrate.

A detailed structural comparison of all sub-sites in AMY1 and AMY2 appears to be an es-sential approach for understanding the differencesin enzymatic properties. This, however, requiresthree-dimensional structures at high resolution ofvarious enzyme/oligo-saccharide complexes. Suchstudies are under way.

The study of barley α-amylase isozymesAMY1 and AMY2 represents a very interestingcase of the collaborative work between structuraldetermination/analysis, biochemistry and molec-ular biology. This couple is a still unique exam-ple of enzymes with very close primary and ter-tiary structure and with such different biochem-ical and biophysical properties. The explanationfor these differences implies an important workbased on structural analyses combined with ra-tional protein engineering experiments in order tounderstand the activity and the specificity of bothisozymes. Finally, AMY1 and AMY2 are two ma-jor candidates in bio-engineering with focus onindustrial and bio-technological applications us-ing genetically modified and improved barley α-amylases.

Acknowledgements

We warmly thank Annette GAJHEDE, Tine E. GOTT-SCHALK and Nathalie JUGE for the preparations of re-combinant AMY1. We furthermore want to acknowl-

68

edge Michel ROTH and Jean-Luc FERRER at the FIPBM30A beamline at the ESRF synchrotron (Greno-ble, France) for precious help and advice during datacollection. This work was financially supported by theEuropean Union Biotech Programme (Framework IVprogramme Biotech II) AGADE “Alpha-Glucan ActiveDesigner Enzymes” (BIO4-980022). Support from theCNRS (Centre National de la Recherche Scientifique)is also gratefully acknowledged.

References

ABE, J., SIDENIUS, U. & SVENSSON, B. 1993. Arginineis essential for the α-amylase inhibitory activityof the α-amylase/subtilisin inhibitor (BASI) frombarley seeds. Biochem. J. 293: 151–155.

AGHAJARI, N., FELLER, G., GERDAY, C. & HASER, R.1998a. Crystal structures of the psychrophilic α-amylase from Alteromonas haloplanctis in its na-tive form and complexed with an inhibitor [pub-lished erratum appears in Protein Sci. 1998. 7:1481]. Protein Sci. 7: 564–572.

AGHAJARI, N., FELLER, G., GERDAY, C. & HASER, R.1998b. Structures of the psychrophilic Alteromonashaloplanctis α-amylase give insights into cold adap-tation at a molecular level. Structure 6: 1503–1516.

AJANDOUZ, E. H., ABE, J., SVENSSON, B. & MAR-CHIS-MOUREN, G. 1992. Barley malt α-amylase.Purification, action pattern, and subsite mappingof isozyme 1 and two members of the isozyme 2subfamily using p-nitrophenylated maltooligosac-charide substrates. Biochim. Biophys. Acta 1159:193–202.

ANDERSEN, J. S., SØGAARD, M., SVENSSON, B. &ROEPSTORFF, P. 1994. Localization of an O-glycosylated site in the recombinant barley α-amylase 1 produced in yeast and correction of theamino acid sequence using matrix-assisted laserdesorption/ionization mass spectrometry of pep-tide mixtures. Biol. Mass Spectrom. 23: 547–554.

ANDRÉ, G., BULÉON, A., HASER, R. & TRAN, V.1999. Amylose chain behavior in an interactingcontext. III. Complete occupancy of the AMY2barley α-amylase cleft and comparison with bio-chemical data. Biopolymers 50: 751–762.

BANNER, D. W., BLOOMER, A. C., PETSKO, G. A.,PHILLIPS, D. C., POGSON, C. I., WILSON, I. A.,CORPRAN, P. H., FURTH, A. J., MILMAN, J. D.,OFFORD, R. E., PRIDDLE, J. D. & WALEY, S.G. 1975. Structure of chicken muscle triose phos-phate isomerase determined crystallographically at2.5 angstrom resolution using amino acid sequencedata. Nature 255: 609–614.

BERTOFT, E., ANDTFOLK, C. & KULP, S. E. 1984. Ef-fect of pH, temperature, and calcium-ions on bar-ley malt α-amylase isoenzymes. J. Inst. Brew. 90:298–302.

BOEL, E., BRADY, L., BRZOZOWSKI, A. M., DERE-WENDA, Z., DODSON, G. G., JENSEN, V. J., PE-TERSEN, S. B., SWIFT, H., THIM, L. & WOLDIKE,

H. F. 1990. Calcium binding in α-amylases: an X-ray diffraction study at 2.1 A resolution of two en-zymes from Aspergillus. Biochemistry 29: 6244–6249.

BRAYER, G. D., LUO, Y. & WITHERS, S. G. 1995. Thestructure of human pancreatic α-amylase at 1.8 Aresolution and comparisons with related enzymes.Protein Sci. 4: 1730–1742.

BRÜNGER, A. T., ADAMS, P. D., CLORE, G. M., DE-LANO, W. L., GROS, P., GROSSE-KUNSTLEVE, R.W., JIANG, J. S., KUSZEWSKI, J., NILGES, M.,PANNU, N. S., READ, R. J., RICE, L. M., SIMON-SON, T. & WARREN, G. L. 1998. Crystallography& NMR system: A new software suite for macro-molecular structure determination. Acta Crystal-logr. D54: 905–921.

BUSH, D. S., STICHER, L., VAN HUYSTEE, R., WAG-NER, D. & JONES, R. L. 1989. The calcium re-quirement for stability and enzymatic activity oftwo isoforms of barley aleurone α-amylase. J. Biol.Chem. 264: 19392–19398.

CCP4 COLLABORATIVE COMPUTATIONAL PROJECT,NUMBER 4. 1994. The CCP4 suite: programs forprotein crystallography. Acta Crystallogr. D50:760–763.

DAUTER, Z., DAUTER, M., BRZOZOWSKI, A. M.,CHRISTENSEN, S., BORCHERT, T. V., BEIER, L.,WILSON, K. S. & DAVIES, G. J. 1999. X-ray struc-ture of Novamyl, the five-domain “maltogenic” α-amylase from Bacillus stearothermophilus: maltoseand acarbose complexes at 1.7 A resolution. Bio-chemistry 38: 8385–8392.

DAVIES, G. J., WILSON, K. S. & HENRISSAT, B. 1997.Nomenclature for sugar-binding subsites in glyco-syl hydrolases. Biochem. J. 321: 557–559.

GIBSON, R. M. & SVENSSON, B. 1987. Identificationof tryptophanyl residues involved in binding of car-bohydrate ligands to barley α-amylase 2. CarlsbergRes. Commun. 52: 373–379.

GOUET, P., COURCELLE, E., STUART, D. I. & METOZ,F. 1999. ESPript: analysis of multiple sequencealignments in PostScript. Bioinformatics 15: 305–308.

JACOBSEN, J. V. & HIGGINS, T. J. V. 1982. Character-ization of the α-amylases synthesized by aleuronelayers of Himalaya barley in response to gibberellic-acid. Plant Physiol. 70: 1647–1653.

JONES, R. L. & JACOBSEN, J. V. 1991. Regulation ofsynthesis and transport of secreted proteins in ce-real aleurone. Int. Rev. Cytol. 126: 49–88.

KADZIOLA, A., ABE, J., SVENSSON, B. & HASER, R.1994. Crystal and molecular structure of barley α-amylase. J. Mol. Biol. 239: 104–121.

KADZIOLA, A., SØGAARD, M., SVENSSON, B. &HASER, R. 1998. Molecular structure of a barley α-amylase-inhibitor complex: implications for starchbinding and catalysis. J. Mol. Biol. 278: 205–217.

LEAH, R. & MUNDY, J. 1989. The bifunctional α-amylase/subtilisin inhibitor of barley: nucleotidesequence and patterns of seed specific expression.Plant Mol. Biol. 12: 673–682.

69

MACGREGOR, A. W. & BALLANCE, D. L. 1980. Hy-drolysis of large and small starch granules fromnormal and waxy barley cultivars by α-amylasesfrom barley malt. Cereal Chem. 57: 397–402.

MACGREGOR, A. W. & MORGAN, J. E. 1986. Hydrol-ysis of barley starch granules by α-amylases frombarley malt. Cereal Foods World 31: 688–693.

MACHIUS, M., DECLERCK, N., HUBER, R. & WIE-GAND, G. 1998. Activation of Bacillus licheni-formis α-amylase through a disorder→ order tran-sition of the substrate-binding site mediated by acalcium-sodium-calcium metal triad. Structure 6:281–292.

MATSUURA, Y., KUSUNOKI, M., HARADA, W. &KAKUDO, M. 1984. Structure and possible cat-alytic residues of Taka-amylase A. J. Biochem. 95:697–702.

MUNDY, J., SVENDSEN, I. & HEJGAARD, J. 1983. Bar-ley αα-amylase/subtilisin inhibitor. Isolation andcharacterization. Carlsberg Res. Commun. 48: 81–90.

NAVAZA, J. 2001. Implementation of molecular re-placement in AMoRe. Acta Crystallogr. D57:1367–1372.

NICHOLLS, A., SHARP, K. A. & HONIG, B. 1991. Pro-tein folding and association: insights from the in-terfacial and thermodynamic properties of hydro-carbons. Proteins 11: 281–296.

QIAN, M., HASER, R. & PAYAN, F. 1993. Structureand molecular model refinement of pig pancreaticα-amylase at 2.1 A resolution. J. Mol. Biol. 231:785–799.

RAMASUBBU, N., PALOTH, V., LUO, Y. G., BRAYER,G. D. & LEVINE, M. J. 1996. Structure of humansalivary α-amylase at 1.6 A resolution: Implica-tions for its role in the oral cavity. Acta Crystallogr.D52: 435–446.

ROBERT, X., GOTTSCHALK, T.E., HASER, R., SVENS-SON, B. & AGHAJARI, N. 2002. Expression, purifi-cation and preliminary crystallographic studies ofα-amylase isozyme 1 from barley seeds. Acta Crys-tallogr. D58: 683–686.

RODENBURG, K. W., JUGE, N., GUO, X. J., SØGAARD,M., CHAIX, J. C. & SVENSSON, B. 1994. DomainB protruding at the third β strand of the α/β bar-rel in barley α-amylase confers distinct isozyme-specific properties. Eur. J. Biochem. 221: 277–284.

RODENBURG, K. W., VALLÉE, F., JUGE, N., AGHA-JARI, N., GUO, X., HASER, R. & SVENSSON, B.2000. Specific inhibition of barley α-amylase 2by barley α-amylase/subtilisin inhibitor dependson charge interactions and can be conferred toisozyme 1 by mutation. Eur. J. Biochem. 267:1019–1029.

ROGERS, J. C. 1985a. Two barley α-amylase gene fam-ilies are regulated differently in aleurone cells. J.Biol. Chem. 260: 3731–3738.

ROGERS, J. C. 1985b. Conserved amino acid sequencedomains in α-amylases from plants, mammals, andbacteria. Biochem. Biophys. Res. Commun. 128:470–476.

ROGERS, J. C. & MILLIMAN, C. 1983. Isolation and se-quence analysis of a barley α-amylase cDNA clone.J. Biol. Chem. 258: 8169–8174.

ROUSSEL, A. & CAMBILLAU, C. 1989. TURBO-FRODO, pp. 77–78. In: Silicon Graphics GeometryPartners Directory (eds) Silicon Graphics, SiliconGraphics Mountain View, CA.

SIDENIUS, U., OLSEN, K., SVENSSON, B. & CHRIS-TENSEN, U. 1995. Stopped-flow kinetic studiesof the reaction of barley α-amylase/subtilisin in-hibitor and the high pI barley α-amylase. FEBSLett. 361: 250–254.

SØGAARD, M., ANDERSEN, J. S., ROEPSTORFF, P. &SVENSSON, B. 1993a. Electrospray mass spectrom-etry characterization of post-translational modifi-cations of barley α-amylase 1 produced in yeast.Bio/technology 11: 1162–1165.

SØGAARD, M., KADZIOLA, A., HASER, R. & SVENS-SON, B. 1993b. Site-directed mutagenesis of histi-dine 93, aspartic acid 180, glutamic acid 205, his-tidine 290, and aspartic acid 291 at the active siteand tryptophan 279 at the raw starch binding sitein barley α-amylase 1. J. Biol. Chem. 268: 22480–22484.

SØGAARD, M. & SVENSSON, B. 1990. Expression ofcDNAs encoding barley α-amylase 1 and 2 in yeastand characterization of the secreted proteins. Gene94: 173–179.

STROBL, S., GOMIS-RÜTH, F. X., MASKOS, K., FRANCK,G., HUBER, R. & GLOCKSHUBER, R. 1997. Theα-amylase from the yellow meal worm: completeprimary structure, crystallization and preliminaryX-ray analysis. FEBS Lett. 409: 109–114.

SVENDSEN, I., HEJGAARD, J. & MUNDY, J. 1986.Complete amino acid sequence of the α-amylase/subtilisin inhibitor from barley. Carlsberg Res.Commun. 51: 43–50.

THOMPSON, J. D., HIGGINS, D. G. & GIBSON, T. J.1994. CLUSTAL W: improving the sensitivity ofprogressive multiple sequence alignment throughsequence weighting, position-specific gap penaltiesand weight matrix choice. Nucleic Acids Res. 22:4673–4680.

VALLÉE, F., KADZIOLA, A., BOURNE, Y., JUY, M.,RODENBURG, K. W., SVENSSON, B. & HASER, R.1998. Barley α-amylase bound to its endogenousprotein inhibitor BASI: crystal structure of thecomplex at 1.9 A resolution. Structure 6: 649–659.

WESELAKE, R. J. & HILL, R. D. 1983. Inhibition ofα-amylase-catalyzed starch granule hydrolysis bycycloheptaamylose. Cereal Chem. 60: 98–101.

Received January 4, 2002Accepted February 20, 2002

70